FTC Opens an Investigation of Anthropic and OpenAI — and Pulls in METR
The agency is drafting its civil investigative demands right now. Also in scope: METR, the very outfit both labs hired to review their own agent incidents.
The agency is drafting its civil investigative demands right now. Also in scope: METR, the very outfit both labs hired to review their own agent incidents.
Google's new frontier model writes up to a million output tokens, where earlier Gemini models stopped at 64,000. Google's own agents used it to rewrite the libgav1 video decoder in safe Rust: 2.7 times faster, identical output.
The Wall Street Journal has the order of events from people who were there: first the agreement to sign, then the questions put to Amodei, then the picture outside the West Wing.
Of every dollar that moves through Amazon's and Google's marketplaces, the two keep about 16 cents. In 2025 that came to roughly $351 million.
Messages sent between 14:00 and 14:59 UTC may not have been saved, Anthropic says.
Claude Desktop jumps from 2.9939.4 to 2.16120.0 the same day and can create a Google Doc straight from the Output picker.
Stripping the refusals out of the model cost Anthropic's own team 2,200 GPU hours and about $4,400. The capabilities survived the edit.
In a footnote OpenAI concedes that Claude Fable 5.1 looks cheaper in its own chart than it really is: on roughly 40 percent of tasks the cost of fallbacks is missing.